As the U.S. approaches the midterms — and both electricity prices and data center opposition become major voter priorities — whether data centers can actually offset utility bills and benefit local grids has become a key political talking point.
At the same time, there’s been a dramatic uptick in the creation of large load tariffs at the state level designed to do exactly that. Over the last two years, utilities have started to converge around certain framework elements to help shield existing ratepayers from the costs and risks of data center growth, like 15-year minimum term lengths, exit fees, and minimum demand charges. Even so, the price data centers will eventually pay to secure grid power is likely to vary dramatically across the country.
According to an analysis conducted by RMI, some utilities are poised to pull in massive sums each year as large hyperscale campuses come online. Portland General Electric, for example, could bring in more than $300 million per year for a 300-megawatt data center, under its large load tariff framework, approved by Oregon’s public utilities commission this spring. This was the most lucrative of all the tariffs that RMI evaluated.
(RMI’s approach to comparing each tariff was to model the total revenue that a generic 300-MW customer operating at a 100% load factor would produce under each.)
PGE’s tariff, mandated by the state legislature, sets minimum billing demand at 90% of contracted capacity, includes terms of up to 30 years depending on project size, and applies a one cent per kilowatt-hour community benefit surcharge for loads over 100 MW. It also includes a “peak growth modifier,” which assigns a greater share of new generation and transmission costs to fast-growing customer classes (read: data centers), plus charges for distribution upgrades needed for specific projects.
Roughly two-thirds of PGE’s modeled tariff revenue, or $195 million, stems from riders and adjustments, with the remaining third attributed to demand and energy charges.
PGE is followed by Dominion, whose tariff will take effect in January, and which RMI estimates would bring in a little over $241 million for the same hypothetical 300-MW data center. Dominion requires large loads to pay 85% of expected transmission and distribution demand, and 60% of generation. It also sets a 14-year contract, and requires loads to post collateral of $1.5 million per MW.
On the other end of RMI’s analysis is Evergy Kansas, which would earn $184 million for the same 300 MW data center. Evergy’s tariff, approved in 2025, includes a 12-year contract term following an optional load ramp-up period of five years, and an 80% minimum demand floor.
The largest driver of the gap between PGE and Evergy’s potential earnings is pass-through riders: Both utilities collect just over $100 million from demand and energy charges, but PGE earns more than $114 million more than Evergy in riders and adjustments. PGE’s state-mandated community benefit surcharge alone adds $26.28 million annually, and its fuel cost adjustment rider adds another $130 million — more than double what Evergy collects through its equivalent fuel rider.
Evergy, for its part, has already signed five massive data centers across its territories in Kansas and Missouri, and is planning to delay the retirement of several coal plants, as well as to build 3.9 gigawatts of new natural gas generation in order to meet the demand.
Applying revenue best practices
RMI’s analysis was introduced as part of a large load tariff proceeding for Xcel Energy in Colorado. Last fall, Colorado ordered its utilities to create a framework for serving large loads while protecting existing ratepayers from the reliability risks and costs.
Xcel’s proposal, currently pending before the state PUC, establishes a new rate class for loads of 50 MW or more, and includes a 15-year minimum contract term, an 80% minimum billing demand floor, and exit fees tied to remaining minimum monthly bills, among other things.
RMI’s modeling was done on behalf of the state’s watchdog organization, the Office of the Utility Consumer Advocate, which is tasked with advocating for consumers when utilities seek to raise rates, and was designed to compare Xcel’s proposed tariffs against an array of its peers.
According to RMI, Xcel’s proposed tariff would yield the lowest annual utility revenue (and potential customer savings) among its peers, recovering just $144.8 million annually. But Xcel’s proposal also includes several separate, optional offerings, including a clean transition tariff and a “speed-to-market pathway.” And RMI found that the way Xcel is proposing to implement these pathways is reducing its potential revenue — and potential savings for existing ratepayers.
Xcel’s version of a clean transition tariff, which allows customers to fund clean energy additions, excludes mature renewable energy sources like wind, solar, and battery storage. That suppresses customer participation, the consumer advocate argued in its own filing, and means Xcel misses out on collecting additional revenue from the most readily available clean energy sources. Other utilities, most notably Nevada Energy, which pioneered the model in partnership with Google and geothermal developer Fervo, allow for a broad mix of clean technologies.
Meanwhile, Xcel’s speed-to-market pathway, which includes faster grid connection for customers that negotiate custom service arrangements, relies on privately negotiated, off-tariff special contracts rather than standardized public tariff rules. That means fast-tracked data centers could potentially negotiate weaker ratepayer protections, the advocate argued — or even bypass some of the tariff’s baseline requirements.
To narrow the gap between Xcel’s potential revenue and that of its peers, the Utility Consumer Advocate argued the utility should open up its clean transition tariff to more forms of clean energy, and bring its fast-track program into the public tariff, so that data centers have to provide grid benefits in order to jump the line. It should also modify its automatic billing rules, so that if a data center is delayed in coming online, or uses less power than projected, ratepayers aren’t left covering the bill. (Currently, the proposal defaults to spreading revenue shortfalls across all customer classes.)
And, Xcel should follow PGE’s lead and add a dedicated community benefit fee, which could help fund low-income energy assistance, home weatherization, and grid repairs in lower-income neighborhoods.
Such charges, the advocate added, aren’t a substitute for a data center’s cost responsibility, but they can help “ensure that new large loads contribute to programs that address local and system impacts.”
